Fault detection and diagnosis (FDD) are essential for enhancing the performance of heating, ventilation, and air conditioning (HVAC) systems, preventing energy waste, and ensuring indoor comfort. However, popular data-driven FDD approaches encounter challenges, such as the lack of high-quality labeled data, poor generalization, and the black-box nature of the models, which hinder their adoption in the market. Moreover, most existing studies only developed and validated their FDD models in simulation environments or laboratory settings, overlooking practical challenges of operational HVAC systems. First, this study focuses on implementing Diagnosis Bayesian networks (DBNs) in real-world settings, specifically, air handling units with heat recovery wheels in a campus building in the Netherlands. DBNs have been proven to be promising solutions with advantages in interpretability, robustness to uncertainties, and flexibility. Second, a Kafka-based framework is introduced for real-time monitoring in HVAC systems, enabling continuous and scalable data processing. Third, a comprehensive diagnosis analysis is conducted using both historical operational data and experimental fault data. The results reveal significant discrepancies between design documents and the actual operation of the HVAC system, and the DBN successfully identifies eight out of nine injected faults during experimentation. Additionally, the results uncover issues such as false positives due to DBN's limitations, inherent system faults, and unexpected HVAC system behaviors triggered by the simulated faults, offering critical insights into the operational challenges and diagnostic potential of DBNs in real-world HVAC systems. These contributions can advance the practical deployment of interpretable and robust FDD tools in building energy systems.

Proper co-optimization of photovoltaic driven air conditioning (PVAC) systems with load flexibility and batteries is pivotal for achieving zero energy buildings (ZEBs). However, practical implementation faces challenges from separate optimization with conflicting objectives, neglect of spatial-temporal occupancy features, and limited consideration of energy, economic, and environmental performance. This study proposes a hierarchical multi-objective co-optimization framework for capacity design and control strategy of the PVAC coupling systems, with the two optimization layers sharing the same multi-objective function. The optimization method balances energy, economic, environmental performance by key metrics including thermal comfort satisfaction ratio (TCSR), grid cumulative action power (GP_total), net present value (NPV) and emission reduction (ER). The optimal capacity optimization of PV and batteries for PVAC systems was solved by the NSGA-II and TOPSIS algorithms. Based on the case study of a multi-functional academic building, the optimization results of the off-grid system and the grid-connected system were calculated under different configuration of PV and battery capacity, and the relationship between the indicators was discussed. The optimization method of off-grid PVAC systems achieves 24 % reduction in PV capacity while maintaining 85.85 % TCSR, 123,800 CNY of NPV, and 167.26 tons of ER. Grid-connected systems with 165.88 kW PV capacity and 71.26 kWh battery capacity can achieve 100 % of TCSR, 1861.8 kW of GP_total, 148,300 CNY of NPV, and 174.70 tons of ER. The study provides an innovative and practical method for capacity design and energy control of PVAC coupling systems to achieve zero energy buildings.

Buildings account for approximately 40% of energy consumption in the European Union and over one-third of energy-related greenhouse gas emissions, with a significant portion attributed to heating, ventilation, and air conditioning (HVAC) systems. Effective fault detection and diagnosis (FDD) are essential for reducing energy waste and lowering maintenance costs in HVAC operations. FDD methods for HVAC systems have been extensively studied and can be broadly classified into two categories: knowledge-based and data-driven approaches. Knowledge-based approaches heavily rely on predefined rules and domain expertise and remain the most widely used in existing HVAC systems. Over the past decade, data-driven FDD approaches have gained popularity. However, data-driven FDD approaches require highquality labelled fault datasets for model training, which can be time-consuming and costly to obtain. To address this challenge, various studies have explored the use of generative adversarial networks (GANs) and other data augmentation techniques to synthesize realistic fault data and improve model performance. Despite these advancements, challenges related to generalization, scalability, and the interpretability of black-box models remain key concerns in the adoption of data-driven FDD approaches. [...]

Digitalization of HVAC piping and instrumentation diagrams (P&IDs) is essential for advancing the intelligent transformation of building systems and the building services industry. This work explores Large Language Models (LLMs) for zero-shot P&ID digitization, focusing on symbol detection. Three LLM-assisted approaches are evaluated. The results show that directly applying LLMs to P&ID digitization is highly challenging. By segmenting P&IDs into local crops and pairing them with the full diagram annotated with bounding boxes to provide global context, the LLM achieves improved symbol recognition. The findings highlight both the promise of LLMs and the need for further refinement to enable reliable HVAC P&ID digitization.

Many buildings suffer from operational inefficiencies, leading to uncomfortable indoor environments, poor air quality, and significant energy waste. Developing automatic fault detection and diagnosis (FDD) tools in building energy systems is essential to mitigate these issues, reducing both energy waste and maintenance costs. Diagnostic Bayesian networks (DBNs), as probabilistic graphical models, offer a promising solution due to their interpretability, robustness to uncertainty, scalability, and flexibility. In this paper, the practical applications of DBNs for FDD in building energy systems are comprehensively reviewed. The generic modeling procedure is systematically examined and summarized, covering problem formulation, structure modeling, parameter modeling, and fault isolation and evaluation. Then, the paper provides insights into DBN modeling objectives, modeling types, diagnostic samples, and modeling software based on the 43 key relevant papers. Furthermore, the paper discusses practical challenges such as sensor configuration, baseline estimation, threshold determination, and expert knowledge integration. Finally, the recommendations are provided to guide further research, aiming to enhance DBN implementation for building energy systems in real-world scenarios, thereby supporting the transformation of the building service industry into a smart sector and ultimately improving building energy performance.

Automated fault detection and diagnostics (FDD) can support building energy performance and predictive maintenance by leveraging the vast amounts of data generated by modern building management systems. Diagnostic Bayesian Networks (DBN) offer a particularly promising approach due to their robustness, flexibility and scalability. However, FDD applications in whole building systems are rare, as they require the integration of different building subsystems, with their own potential faults and symptoms, which increases complexity and makes the resulting DBNs system-specific. In order to overcome these limitations, the 4S3F (four symptoms and three faults) method offers a simplified, adaptable framework for FDD implementation across building systems. In this paper, we implement the 4S3F methodology to a whole-building HVAC system in a case study office building located in the Netherlands. Our methodology uses generic, aggregated representations of individual subsystems within the building, such that FDD methods for specific subcomponents can later be incorporated where available. We first define aggregated building system groups (boiler group, chiller group, hydronic groups, ventilation groups, and end user groups) and subsequently define generic faults that can be detected with the existing sensor infrastructure. This simplified system representation is then used to define a DBN to isolate the most probable system-level faults that lead to building-level symptoms. By focusing on the whole building system, this work aims to provide the groundwork to incorporate occupant feedback and behavior in FDD.

Fault detection and diagnosis (FDD) provides several interrelated benefits, including reducing energy waste, enhanced operational efficiency, and maintaining indoor comfort. The initial step in FDD is to detect deviations from normal or expected operation. However, establishing a reliable baseline can be challenging, especially when there is a lack of sufficient system documents or when complex control strategies are involved. This study investigates three feature selection methods for the baseline estimation: expert knowledge-based, correlation-based, and causality-guided, using heating coil valve control estimation as an example. These methods were tested in an office building in the Netherlands. The results show that while the correlation-based method achieved the best estimation, it may lead to false negatives due to features with reverse causality. This study aims to emphasize the necessity of causal analysis in the baseline estimation to achieve reliable FDD in buildings.

Energy waste in buildings can range from 5% to 30% due to faults and inadequate controls. To effectively mitigate energy waste and reduce maintenance costs, the development of Fault Detection and Diagnosis (FDD) algorithms for building energy systems is crucial. Diagnostic Bayesian Networks (DBNs), as graphical probability models, are particularly useful in scenarios where high-quality data is not always available. While many studies have focused on single fault detection using DBNs, the occurrence of multiple simultaneous faults is common, yet the versatility of DBNs in handling such cases is rarely explored. This study adapts a DBN, initially designed for single fault diagnosis, to perform simultaneous fault diagnosis Experiments were conducted on an air handling unit (AHU) in the Netherlands, using implemented simultaneous faults to test the model. The results suggest that the DBN can detect both single and multiple faults effectively.

The energy matching of PV driven air conditioners is influenced by building load demand and PV generation. Merely increasing energy performance of building or PV capacity separately may improve the energy balance on a large time resolution, the real-time energy mismatching problem is still serious. In this study, a coordinated optimization method of PV capacity, building design, and load flexibility is proposed for improving the real-time energy matching of PVAC system. Then, a methodology integrating data mining method (XG Boost) and parametric simulation was developed to identify the determinant parameters of PV system and building design, exploring feature importance and correlations. The results of XG Boost indicate that the PV capacity, shape factor, and SHGC are the most critical factors. Finally, based on the optimized building design, the PCM layer was applied to improve the real time energy matching. To achieve a goal of 90 % ZEP, the PCM capacity can be decreased by 50.4 % and 62.8 % in Guangzhou and Shanghai in the optimized building. Moreover, the PV capacity can be reduced by 23 % in Guangzhou. The findings of this study provide practical guidance for designing PVAC system coupling with building design and energy storage devices.

This study investigates the diagnostic capabilities of a Diagnostic Bayesian Network (DBN) for air handling unit (AHU) components, particularly focusing on the heat recovery wheel (HRW) and heating coil valve (HCV). Unlike data-driven methods relying heavily on high-quality labeled data, this knowledge-based DBN is more suitable for real-world applications, where labeled faulty and normal data are hard to obtain. Notably, existing studies predominantly concentrate on developing DBN for AHU with recirculated air, neglecting thorough investigations into AHU with HRW, a prevalent system in North Europe and increasingly recommended post-COVID-19 for mitigating viral propagation. This paper presents a DBN setup with expert knowledge for an AHU with HRW, which is evaluated using experimental data from an office building in the Netherlands. The results show that the proposed DBN can successfully diagnose typical faults in HRW and HCV.

Real-time nonintrusive occupancy estimation can maximize the use of existing sensors to infer occupant information in buildings with the advantages of fewer privacy concerns and fewer extra device costs. Recently, many deep learning architectures have proven effective in estimating occupancy directly from raw sensor data. However, some handcrafted features manually extracted from statistical and temporal domains might convey additional information for occupancy estimation. In this study, a novel knowledge fusion network for nonintrusive occupancy estimation is proposed to integrate knowledge from two streams, i.e. automatic knowledge stream from a deep learning architecture and handcrafted knowledge stream from manual feature engineering. Moreover, four different fusion modules are investigated to optimize the design of the fusion network. To verify the effectiveness of the proposed network, experiments are conducted in a dataset from the ASHRAE Global Occupant Behavior Database, which is collected from an office space with records of indoor environment parameters, occupant-building interactions, and contextual information. The results demonstrate the superiority of the proposed fusion network, which outperforms five representative algorithms. Furthermore, the ablation study underscores the benefits of knowledge fusion and occupant-building interaction information, showing that the proposed fusion network can enhance the occupancy estimation accuracy by 3.47 % to 9.24 %.